Google has just announced the release of Gemma, a new collection of open-source language models aimed at making them more accessible to a broader audience of developers and researchers. Gemma is based on the technology used in Google’s previous Gemini models, is offered in two smaller sizes, and includes a range of tools and documentation. […]

Gemma is built on the same research and technology that Google used to create its Gemini models. However, Gemma is designed specifically for developers and researchers who are just getting started with AI. Google states that the models are easy to use and come with a number of tools and resources to help developers get started with it, such as a comprehensive documentation set that covers everything from installing Gemma to fine-tuning it for specific tasks.

The name “Gemma” reflects the Latin word “gemma,” meaning “precious stone,” inspired by Gemini. The collection includes two smaller sizes – Gemma 2B and Gemma 7B – designed with safety and responsible use in mind. These models have been vetted by Google to ensure they are not biased or discriminatory.

Here are some key points about Gemma:

– Gemma is designed for developers and researchers who are new to AI.
– It comes with a comprehensive documentation set for easy implementation.
– Available in two smaller sizes – Gemma 2B and Gemma 7B.
– Designed with safety and responsible use in mind.
– Models can run directly on a developer’s laptop or desktop computer.
– Outperforms far larger machines on important benchmarks.

Google AI hopes that by making Gemma open-source and accessible to a larger variety of developers, it will speed up the creation of new language model applications. These applications can range from developing new text formats, translating languages, writing creative content, to providing useful answers to inquiries. Additionally, Gemma is available worldwide starting today.